[R-sig-ME] repeated measures: levels of grouping factor must be less than number of obs

Brent Pedersen bpederse at gmail.com
Mon Apr 18 17:49:38 CEST 2011


On Mon, Apr 18, 2011 at 9:44 AM, Andrew Miles <rstuff.miles at gmail.com> wrote:
> Just to clarify, when you say repeated measures do you mean that you have
> data from multiple individuals from the same family?  Or do you mean that
> you have data on the same individuals across multiple time points (ex. data
> from person one on Monday, and Tuesday, and Wednesday, etc.)?  The latter is
> what is more traditionally referred to as a repeated measures study; the
> first is simply clustering in the data.
>

yes, I was unclear.
I mean the former: data from multiple individuals in the same family.
*In addition*, each proband subject
may be represented twice--once for each of his/her siblings.


> If the first (i.e. the clustered data), the model you have specified is
> sufficient.  Specifically, it will estimate a variance component that will
> tell you how much variation there is in y between family groups.
>
> Hope that helps.

yes, thanks.

>
> Andrew Miles
>
>
> On Apr 18, 2011, at 11:33 AM, Brent Pedersen wrote:
>
>> hi, I posted this to the main R mailing list a while back and people
>> directed me to
>> the lme4 package and this mailing list so I thought I might get more
>> help here...
>>
>> We have some data, a subset of which is pasted at the end of this message.
>> I am trying to understand how to do repeated measures as our study
>> design consists of a subject (proband) and 1 or 2 siblings.
>>
>> Thus far, our model looks like this--with family_id indicating a
>> sibling relationship:
>>
>>> model = lmer(y ~ concordant + age.proband + age.other + sex.proband +
>>> sex.proband + sex.other + (1| family_id), data=df2)
>>
>> is the above formulation sufficient to capture the repeated measures by
>> family?
>> Next, when I do this with lme4a, I get an error:
>>
>>  Error in lmer(y ~ concordant + age.proband + age.other + sex.proband +  :
>>   number of levels of each grouping factor must be less than number of obs
>>
>> I have seen this thread:
>> http://permalink.gmane.org/gmane.comp.lang.r.lme4.devel/3313
>> but the change does not appear to be included in the latest lme4a
>> package from Rforge, is there another repo where this has been
>> incorporated? or is this an error on my part?
>>
>>
>> thanks,
>> -brent
>>
>> concordant  family_id   external_ref.proband    external_ref.other
>> sex.proband sex.other   age.proband age.other   y(fake)
>> T   58  80015550    80015543    M   F   15  19  1
>> F   58  80015550    80016946    M   F   15  8   2
>> T   54  80015499    80015338    F   F   5   7   3
>> F   54  80015499    80013112    F   M   5   13  4
>> F   22  80012269    80012252    F   F   12  10  5
>> F   22  80012269    80018691    F   M   12  8   5
>>
>> _______________________________________________
>> R-sig-mixed-models at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>




More information about the R-sig-mixed-models mailing list